Can psychology tell you which career will suit you? A century of trying
From paper interest tests to people, things, data and ideas: what a century of career psychology actually tells us.
You answer a questionnaire. A few minutes later, it announces that you should become an architect.
You have never expressed an interest in buildings. You do, however, enjoy solving problems and occasionally own a pencil. Apparently, that settles it.
The ambition behind career tests is much older, and more serious, than their least convincing examples suggest. For over a century, researchers have tried to understand which people will thrive in which kinds of work.
They have also been asking several different questions: who will perform well, who will enjoy the work, who will stay, and who will choose it in the first place. A method can answer one reasonably well while being much less useful for another.
This is a selective history of that research, largely from American vocational psychology: how the methods developed, what they actually measured, and why the evidence gives us something more useful than a single destined occupation.
1909: know the person, know the work
In Choosing a Vocation, published in 1909, Frank Parsons set out three ingredients for a sensible career decision: understanding yourself, understanding occupations, and reasoning about the relationship between them. His account included abilities and interests, but also the requirements, rewards and prospects of different lines of work. Read the original book.
It sounds obvious. It remains surprisingly easy to skip the second part.
A person can spend hours examining their personality while knowing almost nothing about what a transport planner, procurement analyst or occupational therapist actually does.
Parsons offered a systematic counselling method, not a modern experiment proving that a particular score predicts future happiness. That distinction matters when we call an old approach “scientific”.
Still, the architecture was durable: gather evidence about both sides before recommending a match. Our practical reading of it is straightforward. A detailed profile of the applicant cannot rescue a vague description of the job.
The 1920s: compare people with people already doing the job
Edward K. Strong introduced his vocational interest inventory in 1927. The approach associated with it compared a person's pattern of likes and dislikes with patterns found among people in different occupations. The publisher's manual explains the method and its history.
This was an empirical approach: look for answers that distinguish occupational groups, rather than rely entirely on an adviser's intuition about the ideal engineer.
Notice what was being measured. An interest questionnaire asks whether something appeals to you. An ability test asks whether you can do it. Enjoying scientific investigation does not establish that you can already design an experiment.
There is also a general limitation to learning from existing workers. Similarity to people who entered an occupation cannot, on its own, tell us who could flourish there but never had the chance.
That is our caution about interpreting the method, not a claim that every occupational comparison is worthless. Patterns of interest can uncover possibilities. They become restrictive when resemblance is treated as permission to enter.
From 1959: Holland gives interests a structure
John Holland's 1959 theory of vocational choice helped establish a way to describe people and occupational environments within the same framework.
In its familiar later form, the model has six interest areas, abbreviated to RIASEC. The O*NET technical manual describes them:
- Realistic: working with tools, machines, materials or living things.
- Investigative: researching, analysing and explaining.
- Artistic: creating and expressing ideas.
- Social: teaching, helping and supporting.
- Enterprising: persuading, leading and selling.
- Conventional: organising information and following systematic procedures.
The six sit around a hexagon. Neighbouring areas are expected to have more in common than areas further apart. The geometry expresses a proposed relationship between interests; it is not just an attractive way to lay out six labels.
The useful feature is that both a person and an occupation can have a profile across the areas. An interest in research can coexist with an interest in teaching or persuasion.
For a graduate, that is more useful than being handed one permanent type. Two people can both enjoy analysis while wanting quite different amounts of public speaking, practical work or creative freedom.
1982: Prediger looks underneath the hexagon
Dale Prediger asked whether two broader dimensions could explain the pattern behind Holland's categories: people–things and data–ideas.
His 1982 paper examined 27 sets of correlations among Holland's types. These included individual scores and the average scores of occupational groups. He was testing the structure of interests, not following a set of graduates to discover whether two scores could guarantee a good career.
“Data” here means orderly records, information and procedures. “Ideas” means more abstract investigation and creation. It is an older vocabulary, so a modern data scientist could be interested in both.
Consider healthcare. Supporting patients, maintaining equipment, organising records and investigating disease all belong to the same sector. They involve different activities and objects of attention.
That example is why the dimensions remain useful as conversation prompts. “I want to work in healthcare” gives you a setting. Asking which parts of the work appeal gives you something more specific to investigate.
The 1960s: the job has to satisfy the person, too
Alongside the interest models, researchers at Minnesota developed the Theory of Work Adjustment. The 1964 statement was written by René Dawis, George England and Lloyd Lofquist.
Its central distinction is wonderfully practical. The worker must meet the environment's requirements, and the environment must meet the worker's.
The theory calls the person's contentment satisfaction, and the environment's assessment of the person satisfactoriness. The university's account of the framework explains both.
An employer might be delighted with your accurate, dependable work while you are thoroughly bored. You might love a role while struggling to meet its demands. Those are different problems.
The approach also makes room for needs and rewards. Someone can enjoy the activities in a job but dislike its insecurity, supervision or working conditions.
For career guidance, the implication is to ask separately about what someone can contribute and what they need in return. “You would be good at this” leaves half the relationship unexplored.
The 1970s: perhaps we should study the job itself
Richard Hackman and Greg Oldham examined how work design relates to motivation. Their 1976 study tested their model with 658 employees in 62 jobs across seven organisations.
The model focused on skill variety, completing an identifiable piece of work, the significance of that work, autonomy and feedback. It also considered differences in how people respond to jobs with those characteristics.
This helps explain why a title can be a poor description of an experience. Picture two junior analysts: one investigates questions and presents recommendations; the other spends most of the week correcting recurring reporting errors. That is an illustration, but the distinction is exactly the kind job-design research asks us to examine.
Later, Frederick Morgeson and Stephen Humphrey's 2006 Work Design Questionnaire study, involving 540 people in 243 distinct jobs, broadened measurement across task, knowledge, social and contextual characteristics.
These studies concern the nature of work. They do not establish a formula that assigns each graduate to a perfect vacancy.
They do give careers advice a better question: what will this person actually spend Tuesday doing?
Super: a career unfolds alongside the rest of a life
Donald Super developed a different challenge to the idea of a once-and-for-all match. His 1953 paper presented a theory of vocational development. His 1980 life-span, life-space account described careers through the combination and sequence of roles people occupy over a lifetime.
Paid work was one role among others, including being a student, parent or citizen. Those roles could make competing demands on time and commitment.
The practical implication is that “suitable” needs a date attached.
A job involving constant travel could fit one period of someone's life and fit another very badly. The person has not necessarily discovered that their original interests were fraudulent. Their circumstances may have changed.
Super's framework is a theory for understanding development, not a controlled demonstration that everyone follows one timetable. For graduates, its useful challenge is to stop making the first decision carry the weight of every future decision.
You can investigate a promising next step without claiming to have solved your whole working life.
1981–1994: people rule careers out before testing them
Linda Gottfredson's 1981 theory of circumscription and compromise examined how young people narrow their occupational aspirations.
Her account included perceptions of gender and social standing, alongside the compromises made when choices meet practical limits. It was a developmental theory, not proof that every child eliminates occupations in the same sequence.
The question it raises is awkward for any matching system: does “that isn't for me” express a considered dislike, or an option someone has learned not to imagine?
Robert Lent, Steven Brown and Gail Hackett's 1994 social cognitive career theory examined related mechanisms: self-efficacy, outcome expectations and goals.
Self-efficacy means believing you can carry out a particular activity. Outcome expectations concern what you think will happen if you do. These differ from both demonstrated ability and enjoyment.
The authors reviewed evidence and proposed testable relationships between these factors. They also acknowledged the limits of a literature heavily reliant on college and other relatively privileged samples.
For a graduate who says “I'm not a technical person”, a useful follow-up is what that belief rests on: experience, feedback, lack of exposure, or a real dislike of the work.
1999: leave room for things you have not encountered
Kathleen Mitchell, S. Al Levin and John Krumboltz's planned happenstance paper challenged the assumption that career counselling should eliminate chance.
Their proposal was to help people generate, recognise and use unexpected opportunities. This was a counselling framework, not a trial showing that abandoning plans produces better careers.
Our practical interpretation is to make exploration deliberate. Talk to someone doing unfamiliar work. Try a small project. Attend a session outside the sector you have already chosen.
Those activities can produce evidence that another round of introspection cannot. You may discover that the bit you enjoy is interviewing users, not writing the report afterwards.
Exploration still costs time, and sometimes money. A useful approach has to fit the person's circumstances. “Just try everything” is not much help to someone working evenings to cover the rent.
The 2000s onwards: what do the results actually predict?
By this point, we have several plausible ways to describe people and jobs. The next question is how those descriptions relate to outcomes.
A 2002 meta-analysis by Timothy Judge, Daniel Heller and Michael Mount combined 334 correlations from 163 independent samples to examine the Big Five personality traits and overall job satisfaction.
Some traits were associated with satisfaction. But predicting who tends to report greater satisfaction across jobs is a different task from identifying which occupation will suit a particular person. The study did not produce a career allocation table.
Interests have predictive value, too. Christopher Nye and colleagues' 2012 meta-analysis brought together 60 studies and approximately 568 correlations concerning interests and performance. It found relationships with performance and persistence in work and academic settings, with person–environment congruence more predictive of performance than interests alone.
That supports taking interests seriously. It does not turn them into a complete assessment of capability, opportunity or working conditions.
The most useful corrective concerns happiness. Kevin Hoff and colleagues' 2020 meta-analysis examined 105 studies, spanning more than 65 years and 39,602 participants.
The estimated corrected correlation between interest fit and overall job satisfaction was 0.19, with a 95% confidence interval of 0.16 to 0.21.
That is a modest positive relationship. It does not mean a 19% chance of happiness, a 19% improvement in satisfaction or 19% of happiness explained. A correlation describes an association across observations.
The review supports the usefulness of interest fit while limiting how much we should promise from it. Its pooled findings are not a personalised prediction for a UK graduate, and an association does not establish that assigning someone a more closely matched job will cause the predicted improvement.
What a century of research gives a graduate
The evidence leaves us with a richer set of questions than “what type of person are you?”
What activities hold your attention? What can you already do, and what are you willing to learn? What do you need from work? Which environments help you contribute? Which options have you rejected without really encountering them?
Our reading of this history is that a useful career recommendation should make its reasoning inspectable.
“This role involves the investigation you enjoyed, offers structured training and meets your location requirement. We still need to establish how much routine reporting it contains.”
You can do something with that. You can ask questions, compare alternatives and discover where the recommendation is wrong.
A label gives you less to investigate.
The aim should be to find worthwhile possibilities, understand the trade-offs and gather better evidence as you go. Your first job is a decision to make with care. It need not be a diagnosis.
References
Frank Parsons. Choosing a Vocation. 1909. The original counselling framework; a historical method rather than a modern outcome-validation study.
The Myers-Briggs Company. Strong Interest Inventory Manual: introductory extract. Pages 1–2. The instrument publisher explains its 1927 origins and occupational comparisons; this is a later manual, not Strong's original 1927 paper.
John L. Holland. A theory of vocational choice. 1959, pages 35–45. The foundational theory; the familiar RIASEC terminology and instruments developed through later versions.
National Center for O*NET Development. O*NET Interest Profiler Manual. 2021, pages 5–8. Technical documentation for the modern interest model and its measurement, rather than evidence of guaranteed career satisfaction.
Dale J. Prediger. Dimensions underlying Holland's hexagon: Missing link between interests and occupations?. 1982, pages 259–287. Tests the data–ideas and things–people dimensions; structural evidence is distinct from predicting individual outcomes.
René V. Dawis, George W. England and Lloyd H. Lofquist. A Theory of Work Adjustment. 1964. Original Minnesota monograph setting out the framework.
University of Minnesota, Vocational Psychology Research. Theory of Work Adjustment. Institutional explanation of the theory's later formulation and its distinction between satisfaction and satisfactoriness.
J. Richard Hackman and Greg R. Oldham. Motivation through the design of work: Test of a theory. 1976, pages 250–279. Study of 658 employees in 62 jobs across seven organisations; not a graduate career-matching trial.
Frederick P. Morgeson and Stephen E. Humphrey. The Work Design Questionnaire (WDQ): Developing and validating a comprehensive measure for assessing job design and the nature of work. 2006, pages 1321–1339. Measurement study involving 540 incumbents in 243 jobs.
Donald E. Super. A theory of vocational development. 1953, pages 185–190. Foundational developmental theory rather than a fixed occupational assignment method.
Donald E. Super. A life-span, life-space approach to career development. 1980, pages 282–298. Describes careers through changing life roles and commitments.
Linda S. Gottfredson. Circumscription and compromise: A developmental theory of occupational aspirations. 1981, pages 545–579. Original theoretical account of how aspirations narrow; its propositions should not be treated as universal developmental laws.
Robert W. Lent, Steven D. Brown and Gail Hackett. Toward a Unifying Social Cognitive Theory of Career and Academic Interest, Choice, and Performance. 1994, pages 79–122. Theory and evidence review connecting efficacy, expectations, interests and goals, with acknowledged sample limitations.
Kathleen E. Mitchell, S. Al Levin and John D. Krumboltz. Planned Happenstance: Constructing Unexpected Career Opportunities. 1999, pages 115–124. A counselling framework for working with chance, not evidence that unplanned careers necessarily outperform planned ones.
Timothy A. Judge, Daniel Heller and Michael K. Mount. Five-factor model of personality and job satisfaction: A meta-analysis. 2002, pages 530–541. Synthesises 334 correlations from 163 samples; examines overall satisfaction rather than personalised occupational choice.
Christopher D. Nye, Rong Su, James Rounds and Fritz Drasgow. Vocational Interests and Performance: A Quantitative Summary of Over 60 Years of Research. 2012, pages 384–403. Meta-analysis spanning work and academic settings; predictive associations do not establish a causal benefit from a matching service.
Kevin A. Hoff, Q. Chelsea Song, Colin J. M. Wee, Wei Ming Jonathan Phan and James Rounds. Interest fit and job satisfaction: A systematic review and meta-analysis. 2020, article 103503. Across 105 studies and 39,602 participants, the corrected correlation with overall job satisfaction was 0.19; this is not a probability of an individual's happiness.